Early prediction of clinical deterioration using data-driven machine-learning modeling of electronic health records.

OBJECTIVES: To develop and evaluate a high-dimensional, data-driven model to identify patients at high risk of clinical deterioration from routinely collected electronic health record (EHR) data. MATERIALS AND METHODS: In this single-center, retrospective cohort study, 488 patients with single-ventricle and shunt-dependent congenital heart disease <6 months old were admitted to the cardiac intensive care unit before stage 2 palliation between 2014 and 2019. Using machine-learning techniques, we developed the Intensive care Warning Index (I-WIN), which systematically assessed 1028 regularly collected EHR variables (vital signs, medications, laboratory tests, and diagnoses) to identify patients in the cardiac intensive care unit at elevated risk of clinical deterioration. An ensemble of 5 extreme gradient boosting models was developed and validated on 203 cases (130 emergent endotracheal intubations, 34 cardiac arrests requiring cardiopulmonary resuscitation, 10 extracorporeal membrane oxygenation cannulations, and 29 cardiac arrests requiring cardiopulmonary resuscitation onto extracorporeal membrane oxygenation) and 378 control periods from 446 patients. RESULTS: At 4 hours before deterioration, the model achieved an area under the receiver operating characteristic curve of 0.92 (95% confidence interval, 0.84-0.98), 0.881 sensitivity, 0.776 positive predictive value, 0.862 specificity, and 0.571 Brier skill score. Performance remained high at 8 hours before deterioration with 0.815 (0.688-0.921) area under the receiver operating characteristic curve. CONCLUSIONS: I-WIN accurately predicted deterioration events in critically-ill infants with high-risk congenital heart disease up to 8 hours before deterioration, potentially allowing clinicians to target interventions. We propose a paradigm shift from conventional expert consensus-based selection of risk factors to a data-driven, machine-learning methodology for risk prediction. With the increased availability of data capture in EHRs, I-WIN can be extended to broader applications in data-rich environments in critical care.

Thromboelastography Changes of Whole Blood Compared to Blood Component Transfusion in Infant Craniosynostosis Surgery.

ABSTRACT: Surgical treatment of craniosynostosis with cranial vault reconstruction in infants is associated with significant blood loss. The optimal blood management approach is an area of active investigation. Thromboelastography (TEG) was used to examine changes in coagulation after surgical blood loss that was managed by transfusion with either whole blood or blood components. Transfusion type was determined by availability of whole blood from the blood bank.This retrospective study examined differences in posttransfusion TEG maximum amplitude (MA), a measure of the maximum clot strength, for patients transfused with whole blood or blood components. We included all patients less than 24 months old who underwent cranial vault remodeling, received intraoperative transfusions with whole blood or blood components, and had baseline and posttransfusion TEG measured. Whole blood was requested for all patients and was preferentially used when it was available from the American Red Cross.Of 48 eligible patients, 30 received whole blood and 18 received blood components. All patients received an intraoperative antifibrinolytic agent. The posttransfusion MA in the whole blood group was 61.8 mm (IQR 59.1, 64.1) compared to 57.9 mm (IQR 50.5, 60.9) in the blood components group (P = 0.010). There was a greater posttransfusion decrease in MA for patients transfused with blood components (median decrease of 7.7 mm [IQR -3.4, 6.3]) compared with whole blood (median decrease of 2.1 mm [IQR -9.6, 7.5] P < 0.001).Transfusion with blood components was associated with a greater decrease in MA that was likely related to decreased postoperative fibrinogen in this group. Patients who received whole blood had higher postoperative fibrinogen levels.

A Retrospective Study of Neurological Complications in Pediatric Patients With Moyamoya Disease Undergoing General Anesthesia.

BACKGROUND: Moyamoya disease is a condition with potentially devastating and permanent neurological sequelae. Adequate volume status and blood pressure, tight control of carbon dioxide to achieve normocarbia, and providing postoperative analgesia to prevent hyperventilation are typical goals that are used during anesthetic care in these patients. The purpose of this study was to assess postanesthesia neurological complications in moyamoya patients undergoing general anesthesia for imaging studies and surgical procedures excluding neurosurgical revascularization. METHODS: We performed a retrospective cohort study examining moyamoya patients who received general anesthesia for imaging studies and nonneurosurgical-revascularization procedures between January 1, 2001 and December 1, 2016 at our quaternary care pediatric hospital. A general anesthetic encounter was excluded if it occurred within 30 days after a revascularization surgery. The electronic medical records of study patients were analyzed for perioperative management, and neurological outcomes within 30 days of an anesthetic were assessed. RESULTS: A total of 58 patients undergoing 351 anesthesia exposures were included in the study. Three patients experienced neurological complications, which included focal neurological weakness, seizure, and altered mental status. The incidence of complications during anesthesia encounters was 0.85% (3/351) with a 95% confidence interval of 0.28-2.62. CONCLUSIONS: Over a 16-year period at our hospital, 3 children with moyamoya disease who underwent anesthesia for nonneurosurgical-revascularization purposes demonstrated postanesthesia neurological symptoms. The symptoms were consistent with transient ischemic attacks and all resolved without long-term sequelae.

Audit of blood product utilization in the care of injured children.

BACKGROUND: Blood product utilization in injured children is poorly characterized; the decision to prepare products or transfuse patients can be difficult due to a lack of reliable evidence of transfusion needs across pediatric age-groups and injury types. We conducted an audit of transfusion practices in pediatric trauma based on age, injuries, and mechanism of injury. METHODS: We reviewed and cross-referenced blood product transfusion practice data from the trauma registry and the anesthesia transfusion record database at a level 1 pediatric trauma center over a 10-year period. Demographic data, injury severity scores, and survival statistics were obtained from the trauma registry. Transfusion rates are reported separately for hospital admission and for intraoperative transfusions for procedures performed during the first two hospital days. Descriptive statistical analysis was used to compare specific groups based on age, injury type, and mechanism of injury. RESULTS: We report 14 569 trauma admissions of 14 606 patients. The transfusion rate during the admission was 1.56% (227/14 569). 4591 (30.9%) admissions had surgical interventions in first two days of hospitalization with an intraoperative transfusion rate of 2.98%. Patients younger than one year had the highest transfusion rate during admission (2.8%), and the highest transfusion rate during surgical procedures performed in the first two days of the admission (18.87%). Admissions due to vascular injuries had the highest transfusion rates in infancy followed by hollow visceral injuries in adolescents (71.4% and 25%, respectively). Vascular injuries in most age-groups also had high transfusion rates ranging from 11% in 5- to 9-year age-group to 71% in infants. Mechanisms with the highest transfusion rates were firearm wounds in patients older than one year and vehicular accidents for patients younger than one year. CONCLUSIONS: The overall blood product needs in the pediatric trauma population are low (1.56%). Selected populations requiring higher rates of need include infants younger than one year, and children with thoracic and vascular injuries. Understanding transfusion patterns is important to optimize resource allocation.

First-attempt success rate of video laryngoscopy in small infants (VISI): a multicentre, randomised controlled trial.

BACKGROUND: Orotracheal intubation of infants using direct laryngoscopy can be challenging. We aimed to investigate whether video laryngoscopy with a standard blade done by anaesthesia clinicians improves the first-attempt success rate of orotracheal intubation and reduces the risk of complications when compared with direct laryngoscopy. We hypothesised that the first-attempt success rate would be higher with video laryngoscopy than with direct laryngoscopy. METHODS: In this multicentre, parallel group, randomised controlled trial, we recruited infants without difficult airways abnormalities requiring orotracheal intubation in operating theatres at four quaternary children's hospitals in the USA and one in Australia. We randomly assigned patients (1:1) to video laryngoscopy or direct laryngoscopy using random permuted blocks of size 2, 4, and 6, and stratified by site and clinician role. Guardians were masked to group assignment. The primary outcome was the proportion of infants with a successful first attempt at orotracheal intubation. Analysis (modified intention-to-treat [mITT] and per-protocol) used a generalised estimating equation model to account for clustering of patients treated by the same clinician and institution, and adjusted for gestational age, American Society of Anesthesiologists physical status, weight, clinician role, and institution. The trial is registered at ClinicalTrials.gov, NCT03396432. FINDINGS: Between June 4, 2018, and Aug 19, 2019, 564 infants were randomly assigned: 282 (50%) to video laryngoscopy and 282 (50%) to direct laryngoscopy. The mean age of infants was 5·5 months (SD 3·3). 274 infants in the video laryngoscopy group and 278 infants in the direct laryngoscopy group were included in the mITT analysis. In the video laryngoscopy group, 254 (93%) infants were successfully intubated on the first attempt compared with 244 (88%) in the direct laryngoscopy group (adjusted absolute risk difference 5·5% [95% CI 0·7 to 10·3]; p=0·024). Severe complications occurred in four (2%) infants in the video laryngoscopy group compared with 15 (5%) in the direct laryngoscopy group (-3·7% [-6·5 to -0·9]; p=0·0087). Fewer oesophageal intubations occurred in the video laryngoscopy group (n=1 [<1%]) compared with in the direct laryngoscopy group (n=7 [3%]; -2·3 [-4·3 to -0·3]; p=0·028). INTERPRETATION: Among anaesthetised infants, using video laryngoscopy with a standard blade improves the first-attempt success rate and reduces complications. FUNDING: Anaesthesia Patient Safety Foundation, Society for Airway Management, and Karl Storz Endoscopy.

Exhaled nitric oxide measurement before pediatric adenotonsillectomy: A feasibility study.

BACKGROUND: Exhaled nitric oxide (eNO) is a known biomarker for the diagnosis and monitoring of bronchial hyperreactivity in adults and children. AIMS: To investigate the potential role of eNO measurement for predicting perioperative respiratory adverse events in children, we sought to determine its feasibility and acceptability before adenotonsillectomy. METHODS: We attempted eNO testing in children, 4-12 years of age, immediately prior to admission for outpatient adenotonsillectomy. We used correlations between eNO levels and postoperative adverse respiratory events to make sample size predictions for future studies that address the predictability of the device. RESULTS: One hundred and three (53%) of 192 children were able to provide an eNO sample. The success rate increased with age from 23% (9%-38%) at age 4 to over 85% (54%-98%) after age 9. Using the eNO normal value (<20 ppb) as a cutoff, an expected sample size to detect a significant difference between children with and without adverse events is 868, assuming that respiratory adverse events occur in 29% of children. CONCLUSIONS: eNO testing on the day of surgery has limited feasibility in children younger than 7 years of age. The most common reason for failure was inadequate physical performance while interacting with the testing device. The role of this respiratory biomarker in the context of perioperative outcomes for pediatric adenotonsillectomy remains unknown and should be further studied with improved technologies.

Preoperative Fluid Fasting Times and Postinduction Low Blood Pressure in Children: A Retrospective Analysis.

BACKGROUND: Children are required to fast before elective general anesthesia. This study hypothesized that prolonged fasting causes volume depletion that manifests as low blood pressure. This study aimed to assess the association between fluid fasting duration and postinduction low blood pressure. METHODS: A retrospective cohort study was performed of 15,543 anesthetized children without preinduction venous access who underwent elective surgery from 2016 to 2017 at Children's Hospital of Philadelphia. Low blood pressure was defined as systolic blood pressure lower than 2 standard deviations below the mean (approximately the 2.5th percentile) for sex- and age-specific reference values. Two epochs were assessed: epoch 1 was from induction to completion of anesthesia preparation, and epoch 2 was during surgical preparation. RESULTS: In epoch 1, the incidence of low systolic blood pressure was 5.2% (697 of 13,497), and no association was observed with the fluid fasting time groups: less than 4 h (4.6%, 141 of 3,081), 4 to 8 h (6.0%, 219 of 3,652), 8 to 12 h (4.9%, 124 of 2,526), and more than 12 h (5.0%, 213 of 4,238). In epoch 2, the incidence of low systolic blood pressure was 6.9% (889 of 12,917) and varied across the fasting groups: less than 4 h (5.6%, 162 of 2,918), 4 to 8 h (8.1%, 285 of 3,531), 8 to 12 h (5.9%, 143 of 2,423), and more than 12 h (7.4%, 299 of 4,045); after adjusting for confounders, fasting 4 to 8 h (adjusted odds ratio, 1.33; 95% CI, 1.07 to 1.64; P = 0.009) and greater than 12 h (adjusted odds ratio, 1.28; 95% CI, 1.04 to 1.57; P = 0.018) were associated with significantly higher odds of low systolic blood pressure compared with the group who fasted less than 4 h, whereas the increased odds of low systolic blood pressure associated with fasting 8 to 12 h (adjusted odds ratio, 1.11; 95% CI, 0.87 to 1.42; P = 0.391) was nonsignificant. CONCLUSIONS: Longer durations of clear fluid fasting in anesthetized children were associated with increased risk of postinduction low blood pressure during surgical preparation, although this association appeared nonlinear.

Secondhand Smoke exposure and risk of Obstructive Sleep Apnea in Children.

OBJECTIVES: Obstructive sleep apnea (OSA) has a prevalence of 4% in children. Few studies have explored the role of secondhand smoke (SHS) on OSA severity and have shown contradicting results. Most studies have focused on the effect of SHS on snoring. This study explored the association of SHS exposure and OSA severity in children aged 3-18 years. METHODS: This is a retrospective single center IRB-approved study. Electronic Medical Records (EMR) were queried between 1/24/2015 and 1/24/2018 to obtain data on SHS exposure with standard questionnaires from perioperative database. SHS was analyzed as a binary variable and OSA was measured using obstructive apnea hypopnea index (OAHI) from polysomnography (PSG) as a continuous variable. Analyses were done on all children and in those with severe OSA (OAHI≥10/h) as a subgroup. RESULTS: EMR query yielded 101,884 children of whom 3776 had PSG. Limiting baseline PSG in 3-18-year-old and reliable information on SHS yielded 167 analyzable children of whom 70 had severe OSA. Children exposed to SHS had significantly more public insurance than non-exposed (p < 0.0001). Among children with severe OSA, median OAHI was significantly higher in SHS exposed compared to non-exposed (29.0vs.19.5,p = 0.04), but not across all children. In multivariable analysis SHS exposure increased OAHI by 48% in severe OSA subgroup (95%CI: 8%-102%; p = 0.01) when adjusted for race, body mass index, and adjusted household income. CONCLUSION: Children aged 3-18 years with severe OSA who were exposed to SHS were found to have 1.48 increase in odds of OAHI than those without SHS exposure. Results could be limited by retrospective nature of study and EMR tools.

Duration of preoperative clear fluid fasting and peripheral intravenous catheterization in children: A single-center observational cohort study of 9693 patients.

BACKGROUND: Children routinely undergo inhalational induction of general anesthesia. Intravenous line placement typically occurs after induction of anesthesia and can be challenging, particularly in infants and young children. AIMS: We conducted a retrospective observational study to determine whether there was an association between clear liquid fasting time and the number of peripheral intravenous catheter insertion attempts in anesthetized children. The secondary aim was to identify factors associated with multiple attempts to insert intravenous lines. METHODS: After institutional research board approval, we retrieved a data set of all children between 0 months and 18 years who received general anesthesia at our hospital between January 1, 2016, and September 30, 2017. Data included age, gender, weight, race, ASA status, gestational age, number of peripheral intravenous catheter insertion attempts, any assistive device for insertion, and insertion site. Inclusion criteria were mask induction, ASA status 1 or 2, nonemergency, ambulatory surgical procedures, and placement of a single intravenous line during the anesthetic. RESULTS: A total of 9693 patients were included in the study. Of which 8869 patients required one insertion attempt and 824 underwent multiple insertion attempts. 50% of patients in the single insertion attempt group had clear liquid fasting time <6.9 hours compared to 51.8% of patients requiring multiple attempts. Logistic regression model adjusted for age, ASA status, gender, and BMI did not find an association between duration of clear liquid fasting time and rate of multiple insertion attempts for intravenous catheters (OR: 0.99, 95% CI: 0.98-1.01, P = .47). CONCLUSION: Clear liquid fasting time was not associated with multiple insertion attempts for intravenous line insertion in children receiving general anesthesia. Factors such as patient age, ethnicity, time of day of induction of anesthesia, and American Society of Anesthesiologists Physical Status classification show a greater association with the risk of multiple intravenous line insertion attempts.

Development and Usability Testing of the Society for Pediatric Anesthesia Pedi Crisis Mobile Application.

When life-threatening, critical events occur in the operating room, the fast-paced, high-distraction atmosphere often leaves little time to think or deliberate about management options. Success depends on applying a team approach to quickly implement well-rehearsed, systematic, evidence-based assessment and treatment protocols. Mobile devices offer resources for readily accessible, easily updatable information that can be invaluable during perioperative critical events. We developed a mobile device version of the Society for Pediatric Anesthesia 26 Pediatric Crisis paper checklists-the Pedi Crisis 2.0 application-as a resource to support clinician responses to pediatric perioperative life-threatening critical events. Human factors expertise and principles were applied to maximize usability, such as by clustering information into themes that clinicians utilize when accessing cognitive aids during critical events. The electronic environment allowed us to feature optional diagnostic support, optimized navigation, weight-based dosing, critical institution-specific phone numbers pertinent to emergency response, and accessibility for those who want larger font sizes. The design and functionality of the application were optimized for clinician use in real time during actual critical events, and it can also be used for self-study or review. Beta usability testing of the application was conducted with a convenience sample of clinicians at 9 institutions in 2 countries and showed that participants were able to find information quickly and as expected. In addition, clinicians rated the application as slightly above "excellent" overall on an established measure, the Systems Usability Scale, which is a 10-item, widely used and validated Likert scale created to assess usability for a variety of situations. The application can be downloaded, at no cost, for iOS devices from the Apple App Store and for Android devices from the Google Play Store. The processes and principles used in its development are readily applicable to the development of future mobile and electronic applications for the field of anesthesiology.

A Visual Analytics Dashboard to Summarize Serial Anesthesia Records in Pediatric Radiation Treatment.

BACKGROUND: Young children who undergo radiation therapy may require general anesthesia to remain still during weeks of radiation sessions. On a typical day at our hospital, an anesthesia team will care for 10 patients in the radiation therapy suite, and each patient will have multiple prior anesthetic records. Daily review of prior anesthesia records is important to maintain anesthetic consistency and to identify potential improvement, yet our electronic health record (EHR) made such review time-consuming and cumbersome. OBJECTIVES: This article aims to design a visual analytics interface that simultaneously displays data from multiple anesthesia encounters to support clinical consistency in medications and airway management. METHODS: Documentation from the EHR is available in the clinical data warehouse following daily backups. A visual analytics interface was built to aggregate important components of multiple anesthesia encounters in pediatric radiation oncology on a single screen. The application was embedded in the EHR's anesthesia module and updated daily. RESULTS: Each anesthesia encounter was represented by a vertical line with the date at the bottom of the screen. Each vertical line was divided into sections corresponding to the medications, type of airway device, type of radiation oncology procedure, days between treatments, and recovery score and time. Information about the medications, airways, and procedures was shown with icon legends. This layout enabled users to quickly see the key components of multiple anesthetics and make inferences between, for example, the medications used and the recovery score. CONCLUSION: The dashboard provides a high-level summary of all radiation therapy anesthesia records for children receiving recurrent treatments. In this clinical scenario, it is desirable to replicate an optimal anesthetic approach for daily or near-daily treatments or adjust the anesthetic based on observed patterns.